Photonic Integrated Circuit with System on Chip for Sub-Picometric Displacement Sensor – PICSONDE

The first step consists in modeling the optical feedback interferometric signals (OFI) corresponding to the FM channel as well as the FM-to-AM conversion. This modeling step is based on Matlab Simulink in order to simplify its implementation on a digital platform such as FPGAs or SoCs for generating OFI signals. This hardware-in-the-loop implementation aims to facilitate real-time testing of vibration reconstruction algorithms under various regimes and noise levels. The modeling also allows to evaluate the achievable performances by taking into account the various sources of noise and non-linearities from the system.
For the reconstruction of vibrations, algorithms based on the non-uniform sampling method (NUS) coupled with a high frequency modulation of the laser wavelength are developed to reconstruct vibration signals with amplitudes well below half the wavelength of the laser.
This method will be applied to the FM channel of the OFI signals modulated at 10kHz to validate its operating principle and feasibility. Different interpolation methods used by the NUS method will be analyzed in order to identify the most promising one in terms of performance and potential real-time implementation.
To improve the vibration reconstruction, it is also necessary to know the optical feedback factor C. Several methods based on NUS or on machine learning will be implemented and characterized. In fact, the NUS method highlights the presence of discontinuities within the vibration reconstructed displacement that are directly related to C. It is thus possible to assess the value of this coupling factor that can vary over time.
Before performing the FM-to-AM conversion of OFI signals within a dedicated integrated photonic chip (PIC), an experimental setup in free-space configuration is first used to validate the OFI models and reconstruction algorithms. The aim is also to identify the specifications of the future integrated conversion device.
Then a fiber-based version will be developed to identify the future constraints or barriers which might be different from those present in free space. This also allows to characterize different fiber-pigtailed lasers in order to select those presenting the best performances under optical retro-injection.
For the photonic integrated circuits (PICs), a low-loss technology process is chosen. The technology and process will be characterized through the implementation of fundamental components (waveguide, grating coupler, broadband coupler, ...). Then the design of different optical filter architectures (Mach Zehnder, ring resonator, Michelson) as well as their implementation and complete optical characterization will be carried out. Full FM-to-AM conversion PICs will subsequently be implemented and characterized.
Finally, approaches for packaging and connectorizing the PIC to fibers on one hand and the optical sensor with System on Chip on the other hand, will be undertaken to finalize the complete optical sensor.

A complete model developed under Simulink (Matlab) on the phenomenon of optical feedback interferometry (OFI) has been successfully developed including both the amplitude modulated part (AM) of the optical power and the optical frequency (FM) of the signal OFI in the laser diode. The model also takes into account the speckle phenomenon which notably influences the optical coupling coefficient C and the amplitude of the AM interferometric signals. This model was then adapted to be more easily embedded on SoC (Silicon-on-Chip) to be able to emulate OFI signals in real-time.It is currently implemented on a NI acquisition board for real-time generation (1MS/s). This model also allows to validate the different methods used to demodulate the FM OFI channel of the laser using optical filters and subsequently to determine the sensitivity of the device. The study of the different noise sources is also in progress the results of which will be implemented in a future version of the software.
Regarding the optical coupling factor C, a first method based on NUS method exhibited good results with a 5% precision on C under the moderate and strong feedback regimes of the laser. This estimate can be perfomed even if C varies over time. This is the first algorithm offering such a feature. Another method has also been developed, based directly on the amplitude of signals and enabling real-time estimation of C regardless of displacement. In addition, a deep learning approach (Gate Recurrent Units - GRU) is concurrently being developed to render the algorithm more robust. The analysis of the experimental results with this method demonstrates a good correlation with the reference method of estimating C based on the phase unwrapping method, with the added value of being able to estimate values ??of C less than 1. In addition, the impact of the C precision on the reconstruction of the final displacement is also investigated.
The free space FM-to-AM conversion platform has been realized. Preliminary results showed an improvement of one decade compared to the AM channel. The achieved noise spectral density is 10 pm/vHz at 100Hz and below the pmvHz above 5kHz. The platform also validates the FM model developed and allows to test the reconstruction of the displacement using the NUS method with a modulation of the laser up to 10kHz from the FM-to-AM demodulated RIO signals.
Finally, the conversion of the OFI FM channel into optical power modulation using an integrated photonic circuit (PIC) in silicon nitride technology (Mach zehnder or micro-ring resonator) is considered. The sensitivity achieved by the PIC is 1/GHz with a loss of ~1.5dB/cm, sensitivity similar to that obtained with the HCN gas cell. This demonstrates the possibility of being able to integrate this processing part into a PIC and the achieved noise equivalent displacement is 4.9nm for a 1kHz bandwidth.

The aim of PICSONDE consists in attaining a TRL4 for a real-time embedded optical system-on-chip (SoC) vibrometer in order to develop a competitive alternative to piezoelectric and MEMS-based accelerometers for Experimental Modal Analysis employed in the field of predictive maintenance (70 to 80% of the current market in predictive maintenance). Contrary to accelerometer-based approaches, our sensor is completely non-intrusive, can access confined areas while offering both a much larger bandwidth (usually limited to a few tens of Hz to 10 kHz for piezoelectric and MEMS sensors) and dynamic range.
The future prospect is to increase the TRL by improving the robustness and accuracy of the embedded real-time signal processing and the compactness of the sensor through an industrial partnership with ACOEM as an end-user. For instance, in the medium term, the laser diode can be integrated together with the PIC and photodiode. This will result in a very compact architecture where both the laser and the PIC can be thermoregulated simultaneously leading to an interferometer with sub-picometer/vHz performances.
Furthermore, over the longer term, PICSONDE can potentially have an impact in the field of autonomous vehicles by proposing new inertial navigation systems. Subsequent to PICSONDE, the PIC, laser diode, photodiode and the integrated circuit of the entire signal processing chain could be fully integrated together within the same package. Additionally, the proposed integrated interferometer could be used to measure MEMS vibrations with sub-picometric resolution. This can lead to innovative structures for new-generation optical accelerometers and gyroscopes with higher precision and resolution. Consequently, instead of the usual analog front-end used to sense the MEMS capacitive changes, the proposed integrated photonic approach based on an optical interferometer readout circuit can potentially enhance the performances of these devices.
In addition to displacement measurements, our proposed sensor can also be used for refractive index measurement via the effective index variation of a photonic waveguide. This will result in a different optical path and thus a slightly different laser optical frequency which can be tracked by the FM-to-AM conversion scheme using a closed-loop approach based on the lock-in to the half-fringe of OFI signal. In addition, as the sensing is performed on-chip (PIC), the system can be implemented as a Lab-on-Chip where the sample volume necessary could be very small and multi-analysis can be performed simultaneously

International papers:
C. Deleau et al., «Optical Feedback FM-to-AM Conversion With Photonic Integrated Circuits for Displacement Sensing Applications,« in Journal of Lightwave Technology, vol. 42, no. 9, pp. 3446-3453, 1 May1, 2024, doi: 10.1109/JLT.2024.3355048.

O. D. Bernal, H. C. Seat, U. Zabit and F. Surre, «Direct Estimation of the Optical Feedback Factor C From the Amplitude of the Optical Feedback Interferometric Signal,« in IEEE Transactions on Instrumentation and Measurement, vol. 72, pp. 1-7, 2023, Art no. 7006207, doi: 10.1109/TIM.2023.3300431.

H. S. Bazaz, M. M. Fatimah, L. Asim, U. Zabit and O. D. Bernal, «Integration of Zero Crossing Method in a Non-Uniform Sampling System using Optical Feedback Interferometry,« in IEEE Sensors Journal, doi: 10.1109/JSEN.2023.3275702

S. S. Khurshid, W. Hussain, U. Zabit and O. D. Bernal, «Augmentation assisted robust fringe detection on unseen experimental signals applied to optical feedback interferometry using a deep network,« in IEEE Transactions on Instrumentation and Measurement, doi: 10.1109/TIM.2023.3251409

Asra Abid Siddiqui, Usman Zabit, Olivier Bernal. Fringe Detection and Displacement Sensing for Variable Optical Feedback-Based Self-Mixing Interferometry by Using Deep Neural Networks. Sensors, 2022, 22 (24), pp.9831. ?10.3390/s22249831

Bernal, O. D., Zabit, U., Jayat, F., & Bosch, T. (2021). Toward
an Estimation of the Optical Feedback Factor C on the Fly for
Displacement Sensing. Sensors, 21(10), 3528.

International Conferences
Clément Deleau, Han Cheng Seat, Frederic Surre, Olivier Bernal, “Optical Feedback FM-to-AM Conversion with integrated Micro-Ring Resonator for Displacement Sensing Applications”, AIVELA 15th Conference 21-23 june 2023, Ancona, Italy

C. Deleau, T. Apiphatnaphakul, H. C. Seat, F. Surre, U. Zabit,
F. Carcenac, P.-F. Calmon, T. Bosch, O. Bernal, “Towards
Integrated Optical Feedback FM-to-AM Conversion in Silicon
Nitride for Displacement Sensing Applications”.IEEE Sensors 2022 Conference (30 oct – 2 nov 2022)

O. Bernal, H. C. Seat, F. Surre, U. Zabit, C. Deleau, T. Bosch,
“Non-Uniform Sampling Theory applied to FM Channel Optical
Feedback Interferometry for Displacement Sensors”. IEEE Sensors 2022 Conference (30 oct – 2 nov
2022)

PICSONDE focuses on designing and fabricating an embedded sensing system based on optical feedback interferometry (OFI) in a laser diode, thus integrating the light source, the interferometer, and the photodetector within the same package. This proposed OFI-based sensor combines a photonic integrated circuit (PIC) that allows to retrieve the frequency modulated (FM) channel of the OFI signal, with a system-on-chip (SoC) for both data acquisition and signal processing. The sensor resolution targets the OFI quantum limited performance of 0.1 pm/vHz noise power spectrum density over a large bandwidth (up to 100 kHz) for real-time experimental modal analysis applied to predictive maintenance applications in particular. This performance is three orders of magnitude better than existing OFI systems that process the optical output power modulated by OFI, also referred to as the amplitude modulated (AM) channel. For this purpose, four main technological barriers should be resolved: (1) acquire RIO signals with both the highest dynamic range and signal to noise ratio, (2) retrieve the information embedded within highly non-linear RIO signals, (3) assess in real-time the optical feedback factor between the laser and the target, and (4) detect the interferometric fringes in the presence of speckle. To achieve the target resolution, it is required to exploit the FM channel of OFI signals instead of the AM channel to benefit from greatly enhanced noise performances since the OFI signal is no longer limited by the laser shot noise but by its linewidth. The inherent compactness of the PIC will also enhance the system robustness for operation in an industrial environment. In addition, we propose to assimilate the OFI signals to an inherent non-uniform sampling system to be directly processed by an SoC. The laser thus becomes both the sensor and the non-uniform optical data acquisition system. By combining the laser with the PIC and the signal processing SoC, a sensor prototype will be fabricated, and in situ tests will be carried out in collaboration with an industrial end user for predictive maintenance. Nevertheless, this compact high-resolution interferometric sensor can also be used to measure refractive index variations that correspond to optical path changes. Additionally, the sensor can, via the PIC, potentially be exploited to optically probe MEMS structures. This opens a pathway to develop high-resolution (bio)chemical refractometric sensors and optically integrated MEMs sensors.